Dear Joerg,
When I ran the registration between ref_1H and anatomy with standard
FLIRT options
(and 6 DOF) it was fine.
The difficultly was in getting good results from BET given that one
dataset was small FOV
with very thick slices and the other had a very large amount of neck.
In the end, these were
the commands I ran to get a good result:
bet anatomy anatomy_bet -c 80 109 152
bet ref_1H ref_1H_bet -f 0.35 -g 0.28
flirt -in ref_1H_bet -ref anatomy_bet -dof 6 -omat ref2anat.mat -out
ref2anat
Note the -c in bet is helpful for large amounts of neck (locating the
centre in the
centre of the brain) and that the -f and -g options were needed for the
combination
of thick slices and small FOV in the other image.
Try this and see if it also gives you a good result.
All the best,
Mark
Joerg Magerkurth wrote:
>Hi,
>
>following Data:
>
>Highresolution 3D-VRM, t1-mprage, called anatomy
>19 Slices 5mm thickness, t2, called ref_1H
>3 Sclices 5mm thickness, t2, called referenz_1H
>
>task:
>The hi-res image should have the same orientation like the referenz_1H.
>
>2 problems:
>Sometimes the ref_1H has indeed the orientation of the referenz_1H but an in
> plane shift. So a registration of referenz_1H on ref_1H and an inverse
>realign of ref_1H to referenz_1H should solve this.
>The bigger problem is the registration of the the ref_1H with the anatomy.
>With standard-parameters in flirt it doesn't work. So I try to adjust the
>FOV of ref_1H to the anatomy (this was an advise of FSL-list), the result is
>even worse. The only solution which solves the problem in part is to
>registrate the ref_1H to the anatomy without doing bet, cost=mutualinfo and
>limit the search angles to "-searchrx -60 60 -searchry -10 10 -searchrz -20 20".
>Is there a better way to do it? And how I make the registration of
>referenz_1H to ref_1H as good as possible?
>
>I have send the data and the rawdata (dicom) to [log in to unmask] I would
>appreciative for a look at the data.
>
>Best regards Jörg
>
>
|